Construction and Building Materials 47 (2013) 1217–1224

Contents lists available at SciVerse ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier .com/locate /conbui ldmat

Mechanical and fresh properties of high-strength self-compactingconcrete containing class C fly ash

0950-0618/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.conbuildmat.2013.06.015

⇑ Corresponding author. Tel.: +64 3 364 2987x7312, mobile: +64 21 505393;fax: +64 3 364 2758.

E-mail addresses: ms.ashtiani@pg.canterbury.ac.nz, ms.ashtiani81@gmail.com(M. Soleymani Ashtiani), allan.scott@canterbury.ac.nz (A.N. Scott), rajesh.dhakal@canterbury.ac.nz (R.P. Dhakal).

Mohammad Soleymani Ashtiani ⇑, Allan N. Scott, Rajesh P. DhakalDepartment of Civil and Natural Resources Engineering, University of Canterbury, Christchurch 8041, New Zealand

h i g h l i g h t s

� A class C fly ash high-strength self-compacting concrete (HSSCC) mix (100 MPa) was designed.� Benchmark mixes of conventionally vibrated high-strength concretes (CVHSC) were developed.� Mechanical properties of all concrete types were tested after 3, 7, 28 and 90 days curing.� New expressions were devised to predict the properties of HSSCC.� Suitability of current expressions for HSSCC was checked modifications were suggested.

a r t i c l e i n f o

Article history:Received 20 June 2012Received in revised form 29 May 2013Accepted 12 June 2013Available online 5 July 2013

Keywords:High-strengthSelf-compacting concreteClass C fly ashMechanical propertiesFresh properties

a b s t r a c t

In the present study, using the locally available materials in Christchurch, New Zealand, a commerciallyreproducible high-strength self-compacting concrete (HSSCC) mix of 100 MPa compressive strength wasdesigned following the available guidelines for normal-strength self-compacting concrete (NSSCC).Benchmark mixes of conventionally vibrated high-strength concrete (CVHSC) were also designed consid-ering the most important parameters in producing comparable concrete mixes; i.e. similar water-to-bin-der (w/b) ratio and comparable concrete compressive strength. It was found that with an equivalent w/bratio, HSSCC develops considerably higher compressive strength (more than 15 MPa) compared to that ofCVHSC. Therefore, a lower w/b ratio was chosen to reproduce CVHSC mix with strength comparable to theHSSCC mix.

Fresh properties (slump cone, slump flow, J-ring, L-box and V-funnel) and mechanical properties (com-pressive, splitting tensile and flexural strengths as well as modulus of elasticity and shrinkage) of all con-crete types were evaluated at 3, 7, 28, and 90 days. The microstructure of the mixes was assessed bymeans of resistivity, porosity and SEM imaging. New expressions were developed to predict differentcharacteristics of HSSCC and suggestions were made to modify the existing equations where applicable.Finally, experimental results of this study were compared with some of the available codal provisions inorder to assess the applicability of the existing criterion to HSSCC.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

The concept of self-compacting concrete (SCC) was first pro-posed by Hajime Okamura in 1986 as a solution to concrete durabil-ity concerns; however the first SCC prototype was developed in1988 [1–3]. SCC represents one of the most important advances inconcrete technology during the past two decades. Because of itssuperior fresh properties, it flows into a uniform level under theinfluence of gravity with the ability to compact itself by means of

its own weight without the requirement of vibration. Even in themost highly congested reinforced concrete (RC) members, SCC isable to flow free of segregation and de-aerate from large voids.

Due to its unique specifications, SCC may contribute signifi-cantly towards improving the quality of concrete structures. Useof SCC offers several benefits to construction practice such as elim-ination of compaction, shortening of construction time, noisereduction, improved homogeneity, and excellent surface quality.Since the advent of SCC, researchers have investigated its mix de-signs [4], fresh and hardened properties [5] and structural perfor-mance in RC members [6]. In cases where the availableexpressions developed for conventionally vibrated concrete (CVC)were unable to predict the behavior of SCC accurately, suitablemodifications have been made and new expressions have beenproposed by different researchers [5,7].

Table 1Chemical composition of cement and fly ash.

Chemical composition Cement (% by mass) Fly ash (% by mass)

SiO2 20.2 40.1Al2O3 4.35 19.8Fe2O3 2.22 12.2CaO 63.8 14.9MgO – 3.68SO3 2.87 0.68Na2O – 1.57K2O – 0.56TiO2 – 1.37P2O5 – 0.32Loss on ignition – 0.1Total chloride content – 0.002

1218 M. Soleymani Ashtiani et al. / Construction and Building Materials 47 (2013) 1217–1224

In an attempt to produce low cost SCC, Akram et al. [8] incorpo-rated various percentages of bagasse ash as replacement for ce-ment. They found some of the fresh properties of the modifiedSCC not conforming to the guidelines. Therefore, adjustments ofmix proportions as well as the percentage of superplasticizer werenecessary. Overall, out of the twenty-five mixes generated in thelaboratory, only five satisfied the fresh property criterion set bythe EFNARC [9,10]. Thus Akram et al. [8] suggested that the freshproperties of the bagasse ash SCC should be checked to arrive ata proper mix. The compressive strength of the modified concreteis reported to be comparable to that of the control mix after28 days. The cost comparison of the two concrete mixes showedaround 35% lesser expenses in producing one meter cube of theformer compared to the latter with the compressive strength ofboth concrete type above 34 MPa. In an experimental study, Wuet al. [11] investigated the workability of self-compacting light-weight concrete (SCLC) using the overall calculation method withfixed fine and course aggregate contents and incorporating lowweight aggregates (LWA). The workability and uniformity of theSCLC and SCC mixes were validated using fresh property tests, col-umn segregations test and cross-section images. The shear flowvelocity is reported to increase but the resistance to segregationdecreases along with an increase in the binder content.

The material investigations on SCC has gone so far thatresearchers have even started scrutinizing the effects of inclusionof nanoparticles of different types in SCC [12–14]. In a recent study,Nazari et al. [12] investigated the role of SiO2 nanoparticles andground granulated blast furnace slag (GGBFS) on the propertiesof SCC ranging in compressive strength between 15 MPa and82 MPa after 7 and 90 days. They found that the increased GGBFScontent of up to 45% (by weight) could potentially improve thesplitting tensile strength and the pore structure of concrete. How-ever, this would increase the weight loss of the specimens in ther-mogravimetric analysis. In addition they reported that the additionof SiO2 nanoparticles of up to 3% (by weight) could improve thecompressive, split tensile and flexural strengths of concrete dueto formation of more hydrated products. Presence of SiO2 nanopar-ticles could also improve the water permeability of concrete as itacts as a nanofiller for the mix.

While most of the available literature has focused on normal-strength self-compacting concrete (NSSCC) with compressivestrengths ranging between 30 and 80 MPa, little has been reportedon the production and properties of high-strength self-compactingconcrete (HSSCC) with compressive strengths in excess of 80 MPa[5,15,16]. Persson [5] briefly reported some of the current applica-tions of SCC with their mix designs and investigated the concretecompressive strengths in excess of 100, 130, 150 and 170 MPa at7, 28, 90 and 180 days, respectively. However the relevant mix de-signs and fresh properties of these mixes were not reported. Pers-son [5] also suggested new expressions to predict the compressivestrength, modulus of elasticity and shrinkage of SCC.

Dinakar et al. [15] investigated the mechanical properties ofhigh-volume fly ash SCC mixtures with compressive strengthsranging from as low as 8 MPa at 3 days to as high as 103 MPa at180 days. They reported the properties of the constituent materi-als, pertinent mix designs and fresh properties (slump flow, J-ringand V-funnel), as well as the hardened properties (compressiveand splitting tensile strengths and modulus of elasticity) of theinvestigated SCC mixes. Dinakar et al. proposed empirical expres-sions to correlate the splitting tensile strength and modulus ofelasticity of SCC to its compressive strength. Unfortunately forthe higher strength SCC mixes reported by Dinakar et al., some ofthe fresh properties were out of the acceptable ranges prescribedby different standards for SCC [17]. El-Dieb [16] assessed themechanical properties of ultra-high-strength self-compacting con-crete (UHSSCC) incorporating steel fibers (0.08–0.12% and 0.52%

volume fraction) with 90-day cube compressive strength rangingbetween 100 MPa and 150 MPa. The durability and microstructuralcharacteristics of UHSSCC were also investigated. El-Dieb con-cluded that all of the assessed mechanical properties were im-proved by inclusion of steel fibers. Except for the slump flowvalues of the investigated mixes which were stated sporadically,none of the other fresh properties were reported.

For a concrete mix to satisfy the self-compactibility characteris-tics, its fresh properties should be within appropriate ranges [17].However, it can be seen in the available literature on HSSCC that,either the fresh properties have not been investigated or they havenot been sufficiently reported. Furthermore the terms ‘‘high-strength’’ and sometimes ‘‘ultra-high-strength’’ have been useddifferently by different researchers for a wide range of concretecompressive strength. Only limited researches have been con-ducted on handling and properties of HSSCC. Considering theincreasing interest in using SCC during the recent years and its var-ious advantages especially in highly congested RC members, high-rise buildings could possibly be one of its future targets. Howeverdue to the higher demands in the capacity of structural members inhigh-rise buildings (especially in the lower stories), incorporatingSCC of normal strength may not be a preferred option. Neverthe-less before a HSSCC mix can be widely used, more research is re-quired to establish its fresh and mechanical properties so thatthe users have confidence in its performance.

In the present study, a HSSCC mix was developed in the labora-tory following the guidelines given for NSSCC [4]; the same mixwas later proved to be commercially reproducible and used forcasting beam-column joints [18]. For comparison, benchmarkmixes of conventionally vibrated high-strength concrete (CVHSC)were also designed considering the two most important parame-ters used in comparing concrete mix designs; i.e. similar water–binder (w/b) ratio and comparable concrete compressive strength.The fresh and hardened properties of all concrete types were as-sessed in detail and new expressions were proposed to predict dif-ferent characteristics of HSSCC. The microstructure of each mixwas characterized by resistivity, porosity and some SEM imaging.Finally, experimental results of this study were compared withthe available expressions proposed by different researchers andcodal provisions originally developed for CVC and CVHSC in orderto assess their applicability to HSSCC; where necessary, modifica-tions were suggested to the available expressions.

2. Materials

Locally available materials in Christchurch, New Zealand were used in order todesign HSSCC and CVHSC mixes. General Purpose (GP) Cement, Fly ash (class C), anda polycarboxylic ether polymer based superplasticizer (SP) were used. Chemicalcomposition, physical properties of the cement and fly ash are given in Tables 1and 2. The particle size distribution (PSD) of cement and fly ash are provided inFig. 1. An X-ray diffraction (XRD) diagram for major cement components is provided

100

M. Soleymani Ashtiani et al. / Construction and Building Materials 47 (2013) 1217–1224 1219

in Fig. 2. Locally available semi-crushed coarse aggregate (maximum size of 13 mm)and fine aggregate (natural river sand) were used in all concrete mixes. Physicalproperties and grading of aggregates are given in Tables 3 and 4. Potable waterwas used in all concrete mixes.

0

20

40

60

80

0 25 50 75 100 125

Particle Size (microns)

Cement

Fly ash

Fig. 1. PSD of cement and fly ash (Sedigraph 5100 [19] were used in determiningthe PSD).

0

100

200

300

0 20 40 60 80

Cou

nts

Degrees 2-Theta

3CaO.SiO2

2CaO.SiO2

Fig. 2. XRD results for cement.

Table 3Physical properties of aggregates.

Property River sand 13 mm

Specific gravity (g/cm3) 2.61 2.67Bulk density (kg/m3) 1550 1530Void ratio 0.54 0.72Water absorption (%) 0.8 0.8

Table 4Aggregates grading.

Sieve size (mm) River sand (% passing) 13 mm (% passing)

13.20 – 96.1

3. Experimental investigations

Mix design method proposed by Su et al. [4] and guidelines pro-vided by EFNARC [9,10] were used in order to reach an initial mixproportioning for HSSCC and a finalized mix was obtained througha series of laboratory trials. Having fixed the HSSCC mix design andconsidering the fact that the same w/b ratio should produce com-parable compressive strengths, the mix proportion for CVHSC-1was reached at by changing the proportions of coarse and fineaggregates (keeping their total quantity the same as in the HSSCCmix). This way, a comparable mix design with respect to materialquantities and w/b ratio was achieved. Nevertheless, as thestrength development was considerably higher in the HSSCC mix,another mix (CVHSC-2) was designed with a lower w/b ratio in or-der to produce comparable concrete compressive strength. Table 5shows the finalized mix designs for all three concrete types.

A pan mixer of 90 L capacity was used to make the concretemixes. Initially, materials were dry-mixed for about a minute be-fore introducing water and SP to the mix. 80% of the required waterwas added in several stages while mixing, and an additional 1 minof mixing followed. The remaining 20% of water (which was pre-mixed with the required SP) was introduced to the mixer and an-other minute of mixing followed. At this stage, concrete was left forabout a minute in the mixer (without mixing); after which the en-tire mixing procedure was concluded with a final minute ofmixing.

For CVHSC mixes, the only fresh property measured was theslump using the standard slump cone test which led to a slump va-lue of 120 mm for both CVHSC-1 and CVHSC-2 mixes. Fresh prop-erties of HSSCC were assessed using the slump-flow, J-ring, L-box,and V-funnel tests. The results of these tests as well as their typicalacceptable ranges [17] are shown in Table 6.

In order to assess the mechanical properties of the concretemixes, cylindrical specimens of 200 mm height and 100 mm diam-eter were cast to assess the compressive and splitting tensilestrengths and the modulus of elasticity. Beams of 120 � 120 mmcross section and 470 mm length were used to investigate the flex-ural strength. In addition, prisms of 75 � 75 mm cross section and280 mm height were prepared to study the drying shrinkage. TheCVHSC-1 and CVHSC-2 specimens were cast in 3 layers and eachlayer was compacted using a vibrating table operated at2500 rpm for 15 s; whereas the HSSCC specimens were cast with-out using any internal or external vibration. After casting, all spec-imens were covered using suitable materials (steel caps forcylinders and wet burlaps for the beams) to avoid excessive waterevaporation. Specimens were de-moulded 24 h after casting andcured in lime water with a temperature of approximately 20 �C un-til the day of the test. For shrinkage, six standard prisms were castfor each concrete mix, de-moulded after 24 h from casting and keptfor 7 days in the curing tank under water. The prisms were then ta-ken out of the curing tank, surface-dried and first readings for thedrying shrinkage were taken which made the datum for futurereadings. Prisms were kept in a standard shrinkage chamber with

Table 2Physical properties of cement and fly ash.

Property Cement Fly ash

Specific gravity (g/cm3) 3.11 2.55Mean diameter (lm) 45.0 20.5Specific surface area (m2/kg) 367 270

9.50 – 42.74.75 99.3 0.32.36 74.9 –1.18 62.5 –0.60 55.4 –0.30 37.2 –0.15 8.8 –

Pan 0.0 0.0

The grading distribution of aggregates was performed based on the NZS 3111 [20].

Table 5Mix proportions for HSSCC, CVHSC-1 and CVHSC-2.

Material HSSCC (kg/m3) CVHSC-1 (kg/m3) CVHSC-2 (kg/m3)

Coarse aggregate 880 1145 1145Fine aggregate 870 605 695Cement 385 385 385Fly ash 165 165 165Water 165 165 148.5Super-plasticizer 3.58 (0.65%) 1.1 (0.2%) 1.93 (0.35%)

Aggregates were used in saturated surface dried (SSD) condition.

Table 6Fresh properties of HSSCC and accepted values.

Test type Experiment Acceptable range

Slump-flow diameter (mm) 750 600–800Slump-flow T500 (second) 4.2 2–7J-Ring flow diameter (mm) 720 580–780J-Ring height Hin � Hout (mm) 6 0–15L-Box ratio 0.92 0.75–1.0V-Funnel T0 (second) 8 6–12V-Funnel T5 (second)a 9.16 T0–(T0 + 3)

a Same test was repeated after concrete settlement for 5 min.

1220 M. Soleymani Ashtiani et al. / Construction and Building Materials 47 (2013) 1217–1224

a relative humidity (RH) range of 48–52% and a constant tempera-ture of 23 �C. Subsequent readings were taken after 14, 21, 28, 42,56, 70, and 90 days (measured from the casting date).

The microstructure of three mixes was assessed primarilythrough resistivity measurements made from approximately25 mm thick disks cut from the cylinders after 28 days of curing.The specimens were vacuum saturated with tap water and an ACcurrent was applied across the specimens and voltage measured.

There is relatively little apparent difference in microstructurebetween the two CVHSC and one HSSCC mix. Scanning electronmicroscope (SEM) images provided in Fig. 3 show very similarstructure at the both the mm and lm level. Some small macro-scale differences are however evident when comparing the averageresistivity test results of 38, 47 and 40 k Ohm cm for CVHSC-1,CVHSC-2 and HSSCC respectively. The two mixes at a w/b ratio of0.3 showed nearly identical resistivity values while the 0.27 w/bmix had a resistivity approximately 8 k Ohm cm greater. The gen-erally similar mircrostructural performance is to be expected giventhe w/b ratio of 0.3 for CVHSC-1 and HSSCC were identical andCVHSC-2 had at w/b ratio of 0.27. The Fly Ash – GP binder systemwas the same for all three mixes.

Fig. 3. SEM images of the (a) CVHSC-1 w/b 0.3, (b

4. Results and discussions

Specimens were tested following relevant ASTM standards after3, 7, 28, and 90 days of curing for evaluation of compressive [21],splitting tensile [22], and flexural strengths [23] as well as modulusof elasticity [24]. In addition, Australian Standard [25] was used formeasuring drying shrinkage of all concrete samples. As evaluatingthe compressive, splitting tensile and flexural strength testsneeded crushing individual samples, three specimens were testedat each specific age (3, 7, 28, 90 days). However, because of thenon-destructive nature of assessing the modulus of elasticity anddrying shrinkage, the same 3 cylinders and 6 small prisms wererepeatedly used throughout the test, respectively. After each testat a specific age, samples were returned to the curing tank or cli-mate chamber. Table 7 shows the experimental results of the mea-sured mechanical properties for all concrete types. Density of theconcrete mixes was found to be 2470.1, 2466.6 and 2489.2 kg/m3

for HSSCC, CVHSC-1 and CVHSC-2, respectively.The development of the concrete compressive strength vs. time

is shown in Fig. 4 for all concrete types. Although the HSSCC andCVHSC-1 mix proportions were identical even in terms of water/binder (w/b) ratios (except for the proportions of coarse and fineaggregates), the former developed considerably higher compres-sive strength at identical concrete ages. In order to achieve thesame global compressive strength, CVHSC required a lower w/b ra-tio to offset issues such as compaction and segregation. Therefore,in the case of CVHSC-2 and HSSCC, despite having water/binder ra-tios of 0.27 and 0.3 respectively, the strengths were very similar.There are a number of factors which may have contributed to thisbehavior.

� The use of vibration when producing CVHSC may have resultedin some degree of partial segregation; however, properlydesigned HSSCC is manufactured without any vibration; thussegregation is not an issue. Hence, reduced segregation ofHSSCC results in better performance which is demonstratedby the increased strength at each age for a given w/b ratio.� Another important issue is the packing factor [4] which affects

both the final strength and strength gain rate. Packing factor ismainly controlled by the proportioning of coarse and fine aggre-gate in a mix which defines how well different particles come incontact and adhere to each other using the available paste(cementitious material, sand and water). Higher packing factormeans better grading of the coarse and fine aggregates whichresults in better compaction and consequently a higher com-

) CVHSC-2 w/b 0.27, and (c) HSSCC w/b 0.3.

Table 7Mechanical properties all concrete mixes.

Age (day) Compressive strength (MPa) Splitting tensile strength (MPa) Flexural strength (MPa) Modulus of elasticity (GPa)

HSSCC CVHSC HSSCC CVHSC HSSCC CVHSC HSSCC CVHSC

1 2 1 2 1 2 1 2

3 49.4 40.7 50.8 5.1 3.8 5.9 6.5 6.3 7.3 34.8 37.5 36.87 70.1 60.4 69.2 6.1 4.8 6.5 8.7 7.5 9.1 40.2 41.6 40.7

28 88.7 69.4 86.2 6.7 6.2 7.3 8.8 9.4 11.1 42.8 42.4 44.790 101.6 85.7 104.5 8.1 6.2 7.5 11.7 10.1 11.6 44.7 46.2 48.5

30

50

70

90

110

0 20 40 60 80 100

Com

pres

sive

Str

engt

h (M

Pa)

Age (Day)

HSSCC

CVHSC-1

CVHSC-2

EN1992

Proposed

Fig. 4. Strength development vs. time.

M. Soleymani Ashtiani et al. / Construction and Building Materials 47 (2013) 1217–1224 1221

pressive strength. Given the fact that HSSCC had more fineaggregate compared to CVHSC, its packing factor was higherthan the latter.� The use of super-plasticizer (SP) has been shown to have a posi-

tive effect on strength development [26]. However, the increasein strength requires that no segregation occurs after adding SP.In this study the amount of SP used in HSSCC was 3.3 and 1.9times of that provided in CVHSC-1 and CVHSC-2, respectively.

Persson [5] suggested an expression for predicting the concretecompressive strength development vs. time based on the w/b ratio.However as explained before, similar w/b ratio did not result incomparable compressive strengths between HSSCC and CVHSC-1;therefore the proposed expression by Persson could not be appliedto predict the experimental results of this study. For comparison,the experimental results were compared with the model givenby the EN1992 [27] model.

f 0cðtÞ ¼ f 0c;28esð1�ffiffiffiffiffiffiffi28=tp

Þ ð1Þ

In Eq. (1), ‘‘f 0c(t)’’ is the compressive strength (MPa) at the age of t(days), ‘‘f 0c;28’’ is the 28-day compressive strength of concrete(MPa), ‘‘s’’ is the coefficient depending on type of cement, and ‘‘t’’is the age of concrete (days). It should be noted that, the strengthgain trend was similar in all cases and the Eurocode model seemedto provide reasonable prediction of concrete strength gain vs. time.However it appeared that,

(1) As a class C fly ash was used in this study, an ‘‘s’’ factor of0.33 provided the best estimation for the experimentalresults. Note that this ‘‘s’’ factor lied between the ‘‘normaland rapid hardening cements’’ and ‘‘slowly hardeningcements’’ (0.25 and 0.38, respectively) categories suggestedby Eurocode. However, both the start and end points of thegraph seemed to deviate from matching the experimental

results. It is expected that such a divergence between themodel and experimental results was introduced because ofthe fly ash usage in the mix which results in ‘‘low-early’’and ‘‘high-late’’ strength gain.

(2) As the expression given by EN1992 [27] was based on the28-day compressive strength of concrete, prediction ofresults was possible after determining the 28-day strength.This meant that the formula required an input from theactual experimental data which needed about a month time;this is a disadvantage when early age predictions arerequired.

Therefore, a new empirical expression was developed (Eq. (2))to model the experimental results of this study more accurately.This expression has the advantage of being based on the 3-daycompressive strength of concrete rather than the 28-day strength.

f 0cðtÞ ¼ f 0c;3easð1�ffiffiffiffiffi3=tp

Þ ð2Þ

In Eq. (2) ‘‘f 0c(t)’’ is the compressive strength (MPa) at the age of t(days), ‘‘f 0c;3’’ is the 3-day compressive strength of concrete (MPa),‘‘t’’ is the age of concrete (days), ‘‘a’’ is the correcting factor forHSSCC (2.65 in this study), and ‘‘s’’ is the coefficient depending ontype of cement (0.32 using Eq. (2)).

Although tensile strength of concrete is an important character-istic which is used for designing and analyzing strength and ser-viceability of concrete structures, it is relatively difficult to bemeasured due to the problems associated with gripping the spec-imen. Therefore, the splitting tensile and flexural strengths (indi-rect tensile strength) are often used instead of the direct tensilestrength test. In the present study the splitting tensile and flexuralstrengths of all concrete types were assessed using the appropriatespecimens mentioned previously and results are shown in (Fig. 5).

According to Fig. 5a, when the compressive strength was com-parable between concrete mixes (HSSCC and CVHSC-2), for a givenstrength the splitting tensile strength of CVHSC-2 was slightlyhigher than that of HSSCC. One possible explanation for this behav-ior was that the CVHSC-2 had higher coarse aggregate contentwhich helped holding individual sections together. Another expla-nation relates to the quality of paste which was believed to have agreater impact on tensile strength. As the CVHSC-2 mix had a bet-ter paste (lower w/b ratio), it should also have a higher tensilestrength for any given compressive strength. However, it seemedby the time when the 90-day strength (compressive strength in ex-cess of 100 MPa) was reached, the paste was virtually fully hy-drated and the HSSCC developed higher splitting tensile strength.The improvement of the paste was then offset by the slightlyweakened global structure of the CVHSC-2 (possible segregationassociated with vibration). The quality of the aggregate and itsinterface then had a much more important role resulting in an im-proved matrix in HSSCC. More focused testing is required to deter-mine the exact cause of the variation in tensile strength betweenHSSCC and CVHSC-2. However, the improved packing factor andhomogeneity of HSSCC (compared to CVHSC-2) resulted in a moreuniform material (superior compaction without vibration) over the

3

4

5

6

7

8

9

40 50 60 70 80 90 100 110

Split

ting

Ten

sile

Str

engt

h (M

Pa)

Compressive Strength (MPa)

HSSCC

CVHSC-1

CVHSC-2

Felekoglu et al

Dinakar et al

(a)

4

5

6

7

8

9

10

11

12

40 50 60 70 80 90 100 110

Fle

xura

l Str

engt

h (M

Pa)

Compressive Strength (MPa)

HSSCCCVHSC-1CVHSC-2NZS 3101ACI 318Proposed

(b)

Fig. 5. Variation of (a) splitting tensile strength and (b) flexural strength vs. the concrete compressive strength.

1222 M. Soleymani Ashtiani et al. / Construction and Building Materials 47 (2013) 1217–1224

full length of the HSSCC specimens, and the possibility of a weakersection was lower in HSSCC compared to that of CVHSC-2.Although the latter should develop higher splitting tensile strengthat a given age, lower homogeneity and material quality could haveprevented full strength development. Splitting tensile test resultsare compared with the models for HSSCC and SCC proposed byDinakar et al. [15] and Felekoglu et al. [7] (Eqs. (3) and (4), respec-tively) in Fig. 5.

fsp ¼ 0:82ffiffiffiffif 0c

qð3Þ

fsp ¼ 0:43f 00:6c ð4Þ

In these equations, ‘‘fsp’’ is the splitting tensile strength (MPa) and‘‘f 0c ’’ is the compressive strength (MPa). Both models seemed to rea-sonably predict the variations of splitting tensile strength vs. con-crete compressive strength for the HSSCC and CVHSC mixes.

In Fig. 5b, it is evident that at most values of compressivestrength, flexural strength of both CVHSC-1 and CVHSC-2 wasmore than that of HSSCC. This was consistent with the relative val-ues of splitting tensile strengths of the three concrete mixes andcan be explained considering the higher quantity of coarse aggre-gate in both CVHSC mixes.

However compared to the CVHSC mixes, the HSSCC mix devel-oped higher flexural strength between 28 and 90 days of age. Thiscan also be attributed to the same mechanism as explained earlierfor the splitting tensile strength development pattern. In the avail-able literature, no expressions could be found to specifically pre-dict the flexural behavior of HSSCC; therefore the experimentalresults were compared to the analytical models given byNZS3101 [28] and ACI318 [29] (Eqs. (5) and (6), respectively).

fr ¼ 0:6ffiffiffiffif 0c

qð5Þ

fr ¼ 0:62ffiffiffiffif 0c

qð6Þ

In these equations, ‘‘fr’’ is the flexural strength or modulus of rup-ture (MPa) and ‘‘f 0c ’’ is the compressive strength (MPa). Accordingto Fig. 5b, neither NZS3101 [28] nor ACI318 [29] were able to pre-dict the flexural strengths of any of the concrete mixes; in fact theyunderestimated the flexural strengths by a large margin. Therefore,a new expression (Eq. (7)) was proposed by the authors to predictthe flexural strength of HSC which gives a conservative predictionof the experimental results (Fig. 5b).

fr ¼ 0:45ðf 0cÞ2=3 ð7Þ

Modulus of elasticity was also investigated for all concretetypes and compared with the proposed expressions for HSSCCand normal concrete available in the literature such as Gardner[30], Dinakar et al. [15] and Persson [5] (Eqs. (8)–(10) respectively).An illustration of this comparison is given in Fig. 6a.

Ec ¼ 4300ffiffiffiffif 0c

qþ 3500 ð8Þ

Ec ¼ 4180ffiffiffiffif 0c

qð9Þ

Ec ¼ 3750ffiffiffiffif 0c

qð10Þ

In Eqs. (8)–(10) ‘‘Ec’’ is the modulus of elasticity (MPa) and ‘‘f 0c ’’ is thecompressive strength (MPa). According to Fig. 6a, the modulus ofelasticity of HSSCC is slightly lower than that of both CVHSC-1and CVHSC-2 which agrees with the available literature [31]. Thelower modulus of elasticity was attributed to the higher volumeof fine materials in HSSCC compared to both CVHSC mixes; i.e.49.7%, 34.6% and 37.8% of the total aggregate content was naturalsand in HSSCC, CVHSC-1 and CVHSC-2 respectively. It is known thatthe modulus of elasticity of a mix directly correlates to the charac-teristics of the constituent materials. Since there was a higher pro-portion of coarse aggregate in the CVHSC mixes, the accompanyingmodulus of elasticity was also greater than the HSSCC. The effect ofmix proportion on the modulus of elasticity was even evident whencomparing the elastic modulus of CVHSC-1 with a higher proportionof coarse aggregate and CVHSC-2. Despite an increased quality ofthe cement paste, the modulus of elasticity of CVHSC-2 (with w/bratio 0.27) was lower than that of CVHSC-2 (with w/b ratio 0.3).Nevertheless, the higher compressive strength started to offset thisdifference after the CVHSC-2 reached its 28-day strength. Amongstthe selected models only the Gardner’s model [30] was capable ofreasonably predicting the modulus of elasticity of the concretemixes used in this research. Although the expressions proposedby Dinakar et al. [15] and Persson [5] were developed to predictthe elastic modulus of HSSCC, they significantly underestimatedthis property of all concrete mixes.

Finally, shrinkage of the designed concrete mixes was alsoinvestigated following the Australian standard [25] and experi-mental results were compared with the available shrinkage modelsprovided in ACI209 [32] and EN1992 [27] (Eqs. (11) and (12),respectively). Averaged shrinkage results for all concrete types aswell as their comparison with the mentioned models are presentedin Fig. 6b.

30

35

40

45

50

30 50 70 90 110

Mod

ulus

of

Ela

stic

ity

(GP

a)

Compressive Strength (MPa)

HSSCCCVHSC-1CVHSC-2GardnerDinakar et alPersson

(a)0

150

300

450

600

750

7 21 35 49 63 77 91Ave

rage

Dry

ing

Shri

nkag

e (M

icro

Str

ain)

Age (Day)

HSSCCCVHSC-1CVHSC-2ACI-209EN1992EN1992-M

(b)

Fig. 6. (a) modulus of elasticity vs. concrete compressive strength and (b) averaged drying shrinkage vs. time.

M. Soleymani Ashtiani et al. / Construction and Building Materials 47 (2013) 1217–1224 1223

ðeshÞt ¼t

gþ t780csh ð11Þ

ecdðtÞ ¼ð30� 0:21f 0cÞð72e�0:046f 0c þ 75� RHÞðt � tsÞ

t � ts þ bcdð2AcU Þ

2 ð12Þ

In Eqs. (11) and (12), ‘‘(esh)t’’ is the shrinkage strain at any age(micro strains), ‘‘t’’ is the concrete age (day), ‘‘g’’ is 35 for shrink-age after 7 days for moist-cured concrete, ‘‘csh’’ is the product ofapplicable correction factors that are associated with relativehumidity, specimen size, slump, FA percentage, cement content,and air content (1.35 for this study), ‘‘ecd(t)’’ is the shrinkagestrain at any age (micro strains), ‘‘f 0c ’’ is the 28-day compressivestrength of concrete (MPa), ‘‘RH’’ is the relative humidity (%),‘‘ts’’ is the curing time (day), ‘‘bcd’’ is 0.007, ‘‘Ac’’ is the cross-sec-tional area (mm2), and ‘‘U’’ is the perimeter of the member ex-posed to atmosphere (mm).

According to Fig. 6b the drying shrinkage of CVHSC mixes wasslightly less than that of HSSCC; the reason for which was the high-er paste content of the latter. Even between CVHSC-1 and CVHSC-2mixes, the former showed less shrinkage which can be attributedto its lower sand content. ACI209 [32] model seemed to providean acceptable prediction of the experimental shrinkage results asit provided a correction factor which was adjusted to suit the de-signed mixes. This factor takes into account the effect of relativehumidity, specimen size, slump, FA percentage, cement content,and air content. In spite of a close prediction of the final shrinkage(at 90 days), ACI209 [32] model underestimated the early-ageshrinkage of all mixes. EN1992 [27] model had more input param-eters to accommodate different variables of shrinkage test, but itsignificantly underestimated the experimental results of thisstudy. It should however be noted that should the EN1992 [27]model have a correction factor like the one in the ACI209 [32] mod-el, it would have given a much closer prediction of the results.Therefore, authors would like to present a modification to theEN1992 [27] expression (Eq. (13)) in the form of a correcting factorwhich takes into account the effect of slump, FA percentage, ce-ment content, and air content (Fig. 4, EN1992-M).

ecd;modifiedðtÞ ¼ ecdðtÞcsh ð13Þ

where ‘‘ecd,modified(t)’’ is the modified shrinkage at any age (microstrain), ‘‘ecd(t)’’ is the calculated shrinkage from the EN1992 modelat any age (micro strain) and ‘‘csh’’ is the proposed correction factor(2.7 in this study).

5. Conclusions

The following conclusions and remarks are made based on theexperimental results of the present study:

� CVHSC-1 mix was designed with identical material proportionsand w/b ratio to that of HSSCC; except for the proportions ofcoarse and fine aggregates (even the total amount of aggregateswas identical between the two mixes). Although HSSCC andCVHSC-1 had similar w/b ratios, the former developed consider-ably more compressive strength due to its higher materialquality.� CVHSC-2 mix was designed by reducing the w/b ratio and

adjusting the coarse and fine aggregates contents in such away that comparable concrete compressive strengths wereachieved. In spite of the lower w/b ratio (0.27) in CVHSC-2, itdeveloped virtually the same compressive strength comparedto that of HSSCC (with 0.3 w/b ratio).� The higher strength in HSSCC was attributed to the improved

homogeneity and lower possibility of partial segregation result-ing from vibration. A better packing factor in HSSCC also con-tributed to the higher strength development. The use ofhigher quantities of super-plasticizer in HSSCC was alsobelieved to partially take part in its higher compressive strengthcompared to CVHSC-2.� EN1992 model was used to predict the rate of strength-gain vs.

time for all mixes and the model was found inefficient for twomain reasons: (a) the 28-day strength is an essential input forthe EN1992 model which unfavorably delays the strength pre-diction procedure and (b) due to the usage of fly-ash theEN1992 model was unable to predict the low-early-high-latestrength-gain behavior of the mixes.� A new model was proposed based on the 3-day strength of the

mixes to predict more accurately the rate of strength develop-ment in the designed mixes. Note that in the suggested modelthe effect of using fly ash in the concrete mixes was alsoaccounted for by devising appropriate correcting factors. Thesefactors can be calibrated to predict the strength developmentbehavior of other concrete mixes.� At comparable compressive strengths the splitting tensile

strength of both CVHSC mixes proved to be slightly higher thanthat of HSSCC before the age of 28 days. This was attributed tothe higher coarse content of the former compared to the latter.However the HSSCC developed more splitting tensile strengthbetween the age of 28 and 90 days in such a way that at theage of 90 days it reached a higher splitting tensile strength thanthat of CVHSC-2. Better material quality, higher homogeneity

1224 M. Soleymani Ashtiani et al. / Construction and Building Materials 47 (2013) 1217–1224

and lesser void in self-compacting concrete were seemed to bethe possible reasons for such phenomenon. Except for slight dif-ferences, the selected models for SCC and HSSCC (proposed byFelekoglu et al. and Dinakar et al., respectively) were capableof predicting the splitting tensile strength of all concrete types.� The flexural strengths of all three concrete mixes followed

almost the same trend as their splitting tensile strength; i.e.both CVHSC mixes maintained higher flexural strength up tothe age of 28 after which HSSCC overtook and reached a higherflexural strength. As the selected models underestimated theflexural strength of all concrete mixes, a new model was pro-posed and proved to accurately predict the experimental resultsof this study.� At a given compressive strength, the modulus of elasticity of

HSSCC was lower compared to that of CVHSC. This happenedbecause the HSSCC had lower coarse aggregate content com-pared to that of both CVHSC mixes. The selected models pro-posed by Dinakar et al. and Persson for HSSCC,underestimated the elastic modulus of all three concrete types.However, Gardner proposed an expression which showed a veryclose correlation with the experimental results of this study.� Finally, HSSCC seemed to result in a slightly higher drying

shrinkage compared to that of both CVHSC mixes. This wasbelieved to come from the higher volume of paste in HSSCC.The ACI-209 and EN1992 models were used to predict theexperimental results; however the former provided bettermatch for the results of this research. Nevertheless before theconcrete age of 60 days, the ACI209 model could not predictthe shrinkage results accurately. As the EN1992 model wasmore flexible with respect to its input variable, a correcting fac-tor was suggested for the original EN1992 model such a waythat the modified model was capable of accurately predictingthe drying shrinkage results of all three concrete mixes.

The above conclusions are based on laboratory experiments ona reasonable number of samples (to cater for sample-to-samplerandomness) made of three different concrete mixes includingHSSCC. Considering that HSSCC may gain greater attention anduse in the near future due to its exceptional fresh and hardenedproperties, more detailed and focused investigations are requiredin order to provide further confidence-boosting evidences fordesigners to implement HSSCC in RC structures. As the availableliterature shows very limited investigations reported on HSSCC,this study provides a very useful database to plan further investi-gation on this special concrete type.

Acknowledgements

The authors would like to thank James Mackechnie of AlliedConcrete for his assistance in this project. Authors would like to ex-tend their sincere thanks to Tim Perigo, technician of the structureslaboratory at the University of Canterbury, for his continuous sup-port throughout the experimental stage of this study. Authorswould like to acknowledge the funding provided for the projectby the Earthquake Commission (EQC) New Zealand, the Universityof Canterbury and the Department of Civil and Natural ResourcesEngineering.

References

[1] Ozawa K, Maekawa K, Kunishima M, Okamura H. Development of highperformance concrete based on the durability design of concrete structures. In:Proceedings of the 2nd East-Asia and Pacific conference on structuralengineering and, construction (EASEC-2); 1989. p. 445–50.

[2] Okamura H, Maekawa K, Ozawa K. High performance concrete. Tokyo, Gihido;1993.

[3] Maekawa K, Ozawa K. Development of SCC’s prototype. Self-compacting high-performance concrete. Soc, Syst Inst 1999:20–32.

[4] Su N, Hsu K-C, Chai H-W. A simple mix design method for self-compactingconcrete. Cem Concr Res 2001;31(12):1799–807.

[5] Persson B. A comparison between mechanical properties of self-compactingconcrete and the corresponding properties of normal concrete. Cem Concr Res2001;31:193–8.

[6] Hassan AAA, Hossain KMA, Lachemi M. Behavior of full-scale self-consolidatingconcrete beams in shear. Cem Concr Compos 2008;30(7):588–96.

[7] Felekoglu B, Turkel S, Baradan B. Effect of water/cement ratio on the fresh andhardened properties of self-compacting concrete. Build Environ2007;42(4):1795–802.

[8] Akram T, Memon SA, Obaid H. Production of low cost self compacting concreteusing bagasse ash. Constr Build Mater 2009;23(2):703–12.

[9] EFNARC. Specification and guidelines for self-compacting concrete. Europeanfederation of procedures and applicators of specialist products for structures;2002.

[10] EFNARC. The European guidelines for self-compacting concrete specification,production and use. european federation of procedures and applicators ofspecialist products for structures; 2005.

[11] Wu Z, Zhang Y, Zheng J, Ding Y. An experimental study on the workability ofself-compacting lightweight concrete. Constr Build Mater2009;23(5):2087–92.

[12] Nazari A, Riahi S. The role of SiO2 nanoparticles and ground granulated blastfurnace slag admixtures on physical, thermal and mechanical properties of selfcompacting concrete. Mater Sci Eng A 2011;528(4–5):2149–57.

[13] Nazari A, Riahi S. Microstructural, thermal, physical and mechanical behaviorof the self compacting concrete containing SiO2 nanoparticles. Mater Sci Eng A.2010;527(29–30):7663–72.

[14] Nazari A, Riahi S. The effect of aluminium oxide nanoparticles on thecompressive strength and structure of self-compacting concrete. Mag ConcrRes 2012;64(1):71–82.

[15] Dinakar P, Babu KG, Santhanam M. Mechanical properties of high-volume flyash self-compacting concrete mixtures. Struct Concr 2008;9(2):109–16.

[16] El-Dieb AS. Mechanical, durability and microstructural characteristics of ultra-high-strength self-compacting concrete incorporating steel fibers. Mater. Des.2009;30(10):4286–92.

[17] Hwang S-D, Khayat KH, Bonneau O. Performance-based specifications of self-consolidating concrete used in structural applications. ACI Mater J2006;103(2):121–9.

[18] Soleymani Ashtiani M, Dhakal RP, Scott AN. Seismic performance of high-strength self-compacting concrete in reinforced concrete beam-column joints.In: 15th World conference on earthquake engineering. Lisbon, Portugal; 2012.

[19] Sedigraph 5100 Particle Size Analyzer. Mircromeritics. p. 6.[20] NZS3111. NZS3111 Methods of test for water and aggregate for concrete.

Wellington, NZ: Standards Association of New Zealand; 1986.[21] C39/C39M-99 A. ASTM C39/C39M-99 Standard test method for compressive

strength of cylindrical concrete specimens. Annual Book of ASTM Standards;2002.

[22] C496-96 A. ASTM C496-96 Standard test method for splitting tensile strengthof cylindrical concrete specimens. Annual Book of ASTM Standards; 2002.

[23] C78-94 A. ASTM C78-94 Standard test method for flexural strength of concrete(using simple beam with third-point loading). Annual Book of ASTMStandards; 2002.

[24] C469-94 A. ASTM C469–94 Standard test method for static modulus ofelasticity and poisson’s ratio of concrete in copression. Annual Book of ASTMStandards; 2002.

[25] AS1012.13. AS1012.13 Methods of testing concrete – determination of thedrying shrinkage of concrete for samples prepared in the field or in thelaboratory. Standards Australia; 1992. p. 12.

[26] Meddah MS, Zitouni S, Belaabes S. Effect of content and particle sizedistribution of coarse aggregate on the compressive strength of concrete.Constr Build Mater 2010;24:505–12. Copyright 2010, The Institution ofEngineering and Technology.

[27] EN1992-2. EN1992-2 Design of concrete structures. Brussels, Belgium: ComiteEuropeen de Normalisation (CEN); 2005.

[28] NZS3101. NZS3101 Concrete structures standard Parts 1 and 2: the design ofconcrete structures and commentary. Wellington, New Zealand: StandardsNew Zealand; 2006. p. 698.

[29] Committee ACI. Building code requirements for structural concrete andcommentary (ACI 318M-05). Farmington Hills, Mich: American ConcreteInstitute; 2005.

[30] Suksawang N, Nassif HH, Najm HS. Evaluation of mechanical properties forself-consolidating, normal, and high-performance concrete. Trans Re Record2006;1979:36–45.

[31] Domone PL. Self-compacting concrete: an analysis of 11 years of case studies.Cem Concr Compos 2006;28(2):197–208.

[32] 209.2R-08 A. Guide for modeling and calculating shrinkage and creep inhardened concrete. ACI Committee 209; 2008. p. 45.